Phase 1 · Pre-internship hours
Required 500 hours captured at approved partner sites. Gate 1 closes at hours met.
Demo Overview · V1
For the Hospitality Management Faculty Team · University of Arkansas · School of Human Environmental Sciences
NextStage
Internship
Experience
A working demo of an internship operating system. Five screens, one Hybrid substrate, a live AI eligibility classifier. Built for the way you actually run placements, not the way the forms ask you to.
Built for the Hospitality Management Faculty Team
Five-faculty program seat
School of Human Environmental Sciences
Five-faculty program seat
School of Human Environmental Sciences
Built by RTI Enterprises
Tommy Rhoads · Founder + CEO
Today
Tommy Rhoads · Founder + CEO
Today
Page 1 · What this replaces
The problem you described
Six scattered forms. Manual eligibility judgments. A Hours Log that lives on paper or in a different system. Faculty time burned re-reading the same offer letters and chasing the same revision conversations every term.
The honest read. You are not short on rigor. You are short on a connective system. The forms exist. The rubric exists in your head and in the program handbook. What is missing is the layer that turns the rubric into a repeatable decision pipeline and turns the Hours Log from a compliance artifact into a live progress signal.
Where the friction lives today
- Eligibility decisions reconstruct the same rubric from memory every placement, every term.
- Hours Log capture happens after the fact, when a student needs a signature, not when the work was done.
- Faculty review threads through email, paper, and verbal handoffs with no single surface for status.
- Student experience is six forms in five places, with no map of where they are in the program.
- Cross-term continuity resets every fall. New cohort, same six forms, same revision conversations.
1 of 8
Page 2 · What this is
Five screens, one substrate
The NextStage Internship Experience is an operating system, not an app. Five screens share one design substrate and one set of reusable components. Every screen reflects the same canonical case (Mackenzie at Courtyard, August 24 start) so faculty + students see the same picture at the same time.
Screen 01 · Student-facing
Student Cockpit
Mackenzie sees her current status, pre-internship hours met, upcoming events, and Hours Log timing. One screen, no form hunting. Verdict and approver visible.
Screen 02 · Faculty-facing
Faculty Console
Three red queues + KPI tiles + 12-intern grid. The morning pivot for the placement coordinator. Drill in on any name, see the same verdict block the student sees.
Screen 03 · Decision-pipeline
Eligibility Engine
Paste the offer letter and duties description. Classifier returns verdict (internship-level or pre-internship-level), evidence chips, gate status, recommended action. Live AI behind the scenes; sample-data fallback for the demo.
Screen 04 · Pedagogy
How This Connects
Today vs OS comparison. Six forms in five places becomes one pipeline. Built so a department chair, a dean, or a hiring partner can see the change in 90 seconds.
Screen 05 · Journey
Phase Picker
Five-phase journey from pre-internship hours through placement to completion and reflection. Each phase is a click-in, each click-in shows the canonical artifacts that phase produces.
2 of 8
Page 3 · The student + the faculty surface
Same canonical case, two viewpoints
Mackenzie Noffz at Courtyard is the canonical placement that runs through every screen. Faculty see her in the 12-intern grid as Approved. Mackenzie sees the same verdict on her Cockpit. Same data, two surfaces, zero double-entry.
Student Cockpit highlights
Pre-internship hours at 102 percent (gate cleared).
Course hours 0 of 250, opens Aug 24.
Verdict block: internship-level position at Courtyard, approved by Suzanne Rhoads on April 29 after two duties description revisions.
Evidence chips visible: Supervisory rotation, Loyalty intake, Night audit cross-train, Weekly reporting.
Next action: nothing on Mackenzie until August 24 first shift. Hours Log opens automatically.
Faculty Console highlights
KPI tiles: 12 interns active, 3 red queue, 2 watch, 7 clean.
Red queue surface drives the morning work. Each red entry has a named next step + days-open counter.
Pivot from grid to per-student drill-in shows the same verdict block the student sees, plus internal notes that stay faculty-only.
Hover on any cell gives the offer letter + duties text + Eligibility Engine output for that placement.
Why the two surfaces share one substrate. The Hybrid Substrate Spec V1 locks 7 reusable components (KPI tile, progress ring, story card, form input, verdict block, hero card, sidebar). Each screen pulls from the same component library. Adding a sixth screen later means one more file referencing the same substrate, not a fresh design pass.
3 of 8
Page 4 · The decision pipeline
Eligibility Engine, live AI behind it
A faculty member pastes the offer letter and duties description. The classifier reads both, scores against a four-signal rubric, returns a verdict + evidence + recommended action. The classifier sees what is on the page. The system layers in the two eligibility gates (hours met, course status) numerically.
The rubric the classifier runs (verbatim)
Internship-level scope requires at least one of these four signals to be named clearly in the duties description.
- Managerial exposure (shift lead rotation, supervisor shadowing, decision authority on guest issues).
- Analytical or reporting component (recurring reports, dashboards, KPI ownership, trend analysis).
- Cross-functional rotation (multiple departments named, training arc across functions).
- Supervisory shadowing as a recurring touchpoint (not a single observation visit).
A position is pre-internship-level when the duties read as a single-task or single-station job with no managerial, analytical, cross-functional, or supervisory signal. Front desk associate, houseman, server, host by themselves are pre-internship-level. The same job titles can land internship-level when the duties description names the four signals above.
Three sample placements built into the demo
Mackenzie at Courtyard
Internship-level. Evidence chips: Supervisory rotation, Loyalty intake, Night audit cross-train, Weekly reporting. Gates cleared. Action: approve.
Marcus at Pineapple Cafe
Pre-internship-level. Duties read as single-station server work. Action: request revision. Send the supervisor the four-signal language; rerun.
Wyatt at Greene County
Eligibility gate not yet met (hours short). Verdict deferred regardless of scope. Action: surface alternate hour-capture sites; revisit at gate.
How the AI is wired. A Cloudflare Worker sits between the UI and Anthropic. The Worker holds the rubric in its system prompt and forces a structured JSON response. If no API key is configured, the Worker degrades to a keyword-scan fallback so the demo runs unsigned. Zero student data persists anywhere in this stack.
4 of 8
Page 5 · The 90-second story
How this connects to the work you already do
The 90-second view a chair, dean, or hiring partner sees. Today vs OS. Same students, same partners, same compliance requirements. Different connective tissue.
Today · Six forms in five places
- Application form (PDF, email-back).
- Pre-internship hours log (paper or shared sheet).
- Site approval form (PDF, faculty signature).
- Duties description (email thread).
- Hours Log (paper, on-site signature).
- Final report (Word doc, end of term).
- Eligibility judgment is faculty memory + program handbook.
- Status lives in 8 inboxes and 1 paper folder.
OS · One pipeline
- Application + duties land in one form.
- Eligibility Engine runs the rubric, returns verdict + evidence.
- Hours Log opens automatically at term start.
- Faculty Console shows red queues, KPI tiles, 12-intern grid.
- Student Cockpit shows the same verdict the faculty sees.
- Phase Picker shows where every student is on the journey.
- One surface for status. No memory reconstruction.
- Same compliance, fewer chair-on-fire mornings.
What does NOT change
- Faculty are the approvers. The classifier recommends; the faculty member decides.
- The four-signal rubric is the rubric. The OS does not invent new criteria.
- The Hours Log is still the Hours Log. The OS makes it open at the right moment instead of after the fact.
- Student data ownership stays where it sits today. The demo Worker persists nothing.
5 of 8
Page 6 · The journey
Five phases, one journey
The Phase Picker view shows the five phases of the NextStage Internship Experience from a student's standpoint. Each phase has a canonical artifact, a faculty touchpoint, and a hand-off rule.
Phase 1
Pre-internship hours
Phase 2
Site application + duties
Phase 3
Eligibility classification
Phase 4
Placement + Hours Log
Phase 5
Completion + reflection
What each phase produces
Phase 2 · Site application + duties
Single combined form. Site contact + supervisor named. Duties description in standard structure.
Phase 3 · Eligibility classification
Engine runs rubric. Faculty approves, requests revision, or defers. Verdict block visible to student.
Phase 4 · Placement + Hours Log
Hours Log auto-opens at term start. Course hours track 0 of 250. Faculty Console shows live progress.
Phase 5 · Completion + reflection
Final report submitted. Faculty close-out. Cert lifts to Complete. Student carries portfolio artifact forward.
Cross-phase view
Phase Picker click-in shows where the current cohort sits across all five phases. Faculty know what is in motion without reconstructing it every Monday.
6 of 8
Page 7 · Build status
What is built, what comes next
This demo is in working order today. The 5 screens render. The Eligibility Engine returns sample-data verdicts even without an API key. The live AI activates when an API key is wired in. The remaining work is deployment + feedback capture, not new design.
Build phases
Phase 0
Foundation · done
Phase 1
Bake-off · done
Phase 2
Council review · done
Phase 3
5 screens built · done
Phase 4-6
Live AI + Overview + Deploy
Remaining work
- Phase 4 deploy. Cloudflare Worker for the live LLM classifier sits ready on disk. ~8 minutes Tommy hands to fire wrangler deploy + paste the workers.dev URL into the Eligibility Engine UI.
- Phase 5 overview + feedback. The document you are reading is Phase 5. Feedback Capture form lives on the last page of this document.
- Phase 6 deploy + walkthrough. Cloudflare Pages hosts the demo at a Bumpers-branded URL. Walkthrough video recorded. Demo script doc lands in your hands with talking points you use when you show the team.
Ship target: Friday May 29, 2026. That gives the faculty team the rest of this week to walk the live screens, read this document, fill out the 8-question feedback form, and return a read to the program coordinator. Three yeses out of five gets a fall pilot conversation moving.
What this is NOT, yet
- Not a UARK production system. No FERPA review, no UARK SSO, no integration with existing UARK student information systems. This is a demo of the connective layer the Hospitality Management faculty team can pilot if the read across the team supports it.
- Not a finished pilot. The pilot conversation comes after the demo lands and the team has read this document.
- Not a commercial RTI product. This is a working demo built for the Hospitality Management program by RTI Enterprises, intended for the faculty team to read, react to, and decide whether to take into a real pilot.
7 of 8
Page 8 · Feedback capture
What would you change?
Structured questions, not free-text noise. The goal is to harvest concrete feedback Tommy can fold into the next iteration before the demo URL goes to your team. Answer the questions that you have a clear answer to; skip the rest. Hand the form back when you are done.
Q1Which screen most resonates with the work you do today?
Student Cockpit
Faculty Console
Eligibility Engine
How This Connects
Phase Picker
Q2Which screen would save you the most time tomorrow if it ran live?
Choose one. Same options as Q1.
Q3What is missing from the eligibility classifier that you would need to trust it?
Up to three short answers. Trust gaps, not feature wishes.
Q4Is the Mackenzie case representative of typical placements you process?
Yes
Mostly
No
One line, why.
Q5What is the smallest piece of this you could pilot in your current term?
Hours Log
Eligibility Engine
Faculty Console
How This Connects
Phase Picker
None of these
Q6If we built this for UARK production, what would the hardest internal political block be?
Up to three short answers. Real blockers, not generic risk language.
Q7One sentence. What did this demo unlock in your head that the existing system has not?
8 of 8
NextStage Internship Experience · Demo Overview V1
Built by Tommy Rhoads, RTI Enterprises · for the Hospitality Management faculty team, University of Arkansas · School of Human Environmental Sciences
Today · Em-dash zero · Ship target Friday May 29